Thesis Respository

I think I’m finding the thesis process strange because it clashes quite violently with how I was taught to engage long-form projects, particularly creative ones.

Writing is the core of a thesis, and not necessarily because the output will be prose, but rather because writing is the only way to digest the ungodly volume of information flooding into you over the span of the project. So - I enjoy a capstone when it means I can spend a large part of it meandering through half-baked thoughts and dead-end ideas. I feel that as long as I make it a point to eek out a page or two ever couple of days, the whole thing seems to fall into place by the end.

I bring this up because, structurally, I think my process will look somewhat like this:

Flow Chart

This might be a bit of a counterintuitive proposition given that, at the moment, most of my direction for thesis seems to be towards code. That said, I think code without context is fairly boring, and I’m hoping this structure will remedy it.

Storage

I need to come up with a better way of nicely “displaying” information about what I’ve read. Again, this is somewhat foreign to me: I’m used to reading, accumulating said reading in notes, references, and loose knowledge maps, and then putting it all together in one final delivery.

This idea that our sources are going to be scrutinized as we go, or that our repository of sources has to be open is, if I’m brutally honest, something I might keep up for the first week or two and then abandon.

The other issue with this has to do with my current knowledge management system - I’ve grown quite fond of my self-hosted SilverBullet repo, which has become a second-nature way to organize my thoughts and jot down lists of sources, as well as link things together.

In a pinch, I might go for something like Notion.

Mindspace

I came to ITP with a fairly clear idea of what I wanted to explore, and what I was interested in at a technical level. This was at the heels of my residency at the Recurse Center, where I focused on the development of Domain-Specific Languages and micro-frameworks. I have always been interested in the sort of utopian elegance that early computer technology had: languages like Lisp, early networking protocols, and now-forgotten interfaces (like project Xanadu) all call out to me as very rich fields of exploration. As of late I’ve become particularly interested in the idea of “non-hierarchical” data structures, which were also attempted (most famously by BeOS) but fell out of favor due to a variety of reasons (both social and technical).

The big problem with a lot of these early visions of technology was a mismatch in complexity: while the systems were based on simple, elegant rules, they in fact depended on complex and significant volumes of data in order to function properly. Take the non-hierarchical file system for example: every file must be tagged, so that instead of “paths”, tags can be used to filter information and create “flexible” hierarchies. What this meant in practice was that someone had to manually input and update tags for every single piece of content in the filesystem. For obvious reasons, this was a no-go.

Something has changed: with the advent of AI, we now have tools that allow us to extract semantic meaning from pretty much anything we create. More importantly, through things like Embeddings, we are able to create geographies of meaning that automatically cluster information semantically. While I’m a sincere believer that AI is ushering - live - the downfall of human civilization, I also think that we shouldn’t really throw out the baby with the bathwater. Embeddings, and semantic extraction, might be what allows us to revisit now-forgotten paradigms in computer technology and HCI, and make them viable models for thought and creation today.

So; I’m quite clear on what I want to explore for my thesis, at a technical level. I’ve started prototyping a system that will hopefully allow me to validate a lot of this thought. But, I’m still working on defining a way of approaching the conceptual side of things - the “writing”. So far, I think I will:

My hope is that this will buttress and inform the choices I make around the priorities of the system, as well as the framework for presenting the project.

Of the projects I saw presented last year, I think the one that struck me as most interesting was Zongze’s, in that it unmasked and reinterpreted the rigid data structures underlying visual representation of online content.

../